Developing a network-based encoding model of motor cortex during natural behavior of the unconstrained marmoset

在不受约束的狨猴自然行为过程中开发基于网络的运动皮层编码模型

• 批准号: 10263941
• 负责人: Dalton D Moore
• 金额: $ 4.6万
• 依托单位: UNIVERSITY OF CHICAGO
• 依托单位国家: 美国
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-09-30 至 2022-09-29
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10263941
• 关键词: Accounting; Algorithms; Behavior; Callithrix; Callithrix jacchus jacchus; Complex; Computer Analysis; Cortical Column; Coupling; Data; Dimensions; Entropy; Experimental Designs; Foundations; Goals; Individual; Interneurons; Label; Lead; Linear Models; Locomotion; Mentorship; Modeling; Motor; Motor Cortex; Movement; Nervous system structure; Network-based; Neurons; Neurosciences; Organism; Output; Pattern; Population; Population Dynamics; Primates; Property; Recording of previous events; Role; Science; Sensory; Shapes; Spinal Cord; Structure; Techniques; Thalamic structure; Time; Training; Upper limb movement; Weight; Work; area striata; base; brain machine interface; improved; insight; kinematics; motor behavior; motor control; musician; novel; predicting response; predictive modeling; relating to nervous system; sound; vector; visual stimulus

Project Summary Any interaction between an organism and a static external world requires a motor behavior, from basic functions such as locomotion to the precise, highly-trained movements performed by athletes and musicians. As a critical node in the nervous system involved in voluntary motor control, the primary motor cortex (M1) provides the opportunity to investigate neuronal computations and their outputs at the final stage of cortical processing before movement. Studies of primate reaching that consider features of neuronal activity beyond population vectors of single-neuron tuning properties often emphasize the role of structured population dynamics in producing motor behavior. However, the dimensionality reduction techniques used to characterize these dynamics discard all information regarding the identity, role, and tuning properties of single neurons, thereby hindering efforts to understand the function of individual neurons within the context of the population. In order to integrate the neuron-centric and network perspectives on computations in motor cortex, encoding models must be developed that can predict fine-timing spike activity by accounting for movement kinematics, layer-specific properties, and population activity. Network science – the study of complex networks – provides a path toward this goal. Network activity will be summarized by functional networks (FNs) that maintain neuron- specific labels while simultaneously capturing all pairwise correlations between neurons. We will use FNs to investigate the role of network interactions conditioned on neuron-specific features, particularly cortical depth and whether it responds preferentially during particular movements (tuned) or not (untuned). I propose the use of unrestrained common marmosets to study single-neuron and network-based representations of naturalistic motor behaviors. I will model trajectory encoding of single neurons in deep and superficial layers, quantify functional connections between neurons within a cortical column and across the cortical sheet, and examine the role of tuned and untuned neurons in shaping the encoding properties of functionally connected units. By leveraging a model that combines temporally precise kinematic encoding with quantified network interactions, we can place single neuron properties in the context of population dynamics and gain insight into the activity patterns that produce natural motor behaviors. Furthermore, the study of unconstrained behavior will produce an improved model that can account for complex, naturalistic movements. This has implications for brain- machine interface control algorithms, where a deeper understanding of the encoding properties and network interactions that produce a rich motor repertoire may provide the foundation for algorithms that achieve dexterous movement in a long-term, unconstrained setting. I seek to accomplish these goals and gain valuable training in sound experimental design, computational analysis, and interpretation of encoding models with the guidance of my sponsor Dr. Nicho Hatsopoulos and collaborator Dr. Jason MacLean, who provide complementary expertise in motor and network neuroscience and a history of superb mentorship.

项目摘要 生物体与静态外部世界之间的任何相互作用都需要运动行为，从基本的 从运动到运动员和音乐家所做的精确、训练有素的动作。 初级运动皮层（M1）是参与自主运动控制的神经系统的关键节点， 提供了机会，研究神经元的计算和他们的输出在最后阶段的皮层 运动前的处理。灵长类动物的研究，考虑了神经元活动的特征， 单神经元调谐特性的群体向量通常强调结构化群体的作用 产生运动行为的动力学。然而，用于表征的降维技术 这些动力学丢弃了关于单个神经元的身份、角色和调谐特性的所有信息， 从而阻碍了在群体环境中理解单个神经元的功能的努力。 为了整合运动皮层计算的神经元中心和网络观点，编码 必须开发能够通过考虑运动运动学来预测精细定时尖峰活动的模型， 层特异性性质和群体活性。网络科学-复杂网络的研究-提供 一条通往这一目标的道路。网络活动将由维持神经元功能的功能网络（FN）总结， 特定的标签，同时捕获神经元之间的所有成对相关性。我们将使用FN来 研究以神经元特异性特征为条件的网络相互作用的作用，特别是皮层深度 以及它在特定运动期间是否优先响应（调谐）或不（未调谐）。我建议使用 研究单神经元和基于网络的自然表达， 运动行为我将模拟深层和浅层单个神经元的轨迹编码， 皮质柱内和皮质片上神经元之间的功能连接，并检查 调谐和未调谐神经元在塑造功能连接单元的编码特性中的作用。通过 利用将时间上精确的运动学编码与量化的网络交互相结合的模型， 我们可以把单个神经元的特性放在种群动力学的背景下， 产生自然运动行为的模式。此外，对无约束行为的研究将产生 一个改进的模型，可以解释复杂的，自然的运动。这对大脑有影响- 机器接口控制算法，在这里更深入地了解编码特性和网络 产生丰富的运动功能的相互作用可以为实现以下目标的算法提供基础： 在长期的、不受约束的环境中的灵巧运动。我努力实现这些目标， 训练在健全的实验设计，计算分析，并与编码模型的解释 我的赞助商Nicho Hatsopoulos博士和合作者Jason MacLean博士的指导，他们提供 在运动和网络神经科学方面的互补专业知识和高超的指导历史。